Abstract
A noticeable topic to be pursued in the field of on-board real-time data processing is the influence of the modulation transfer function (MTF) of the image acquisition system on the lossless compressibility of raw (that is, uncalibrated) hyperspectral data. Actually, notwithstanding the system device is constrained by several design and manufacturing requirements, the impact of the on-board MTF on the performance of data compressors is becoming remarkable. In particular, the aim of reducing both transmission bandwidth/power and mass storage can be efficiently pursued. Such an analysis is expected to be useful especially for systems employed in mini-satellites, whose payload must be compact and light. From this perspective, this paper investigates the performance of a typical imaging system that acquires low/medium-spatial-resolution images, by considering high-resolution reference data, which simulate the real scene to be imaged. To this end, standard Consultative Committee for Space Data Systems (CCSDS) Aviris 2006 data have been chosen, due to their spatial resolution of 17 m, which is adequate to be a reference for simulated data whose spatial resolution is foreseen between 50 and 150 m. MTF requirements are usually provided based on the cut-off value of the amplitude at the Nyquist frequency, which is defined as a half of the sampling frequency. Typically, a cut-off value between 0 . 2 and 0 . 3 ensures that a sufficient amount of information is delivered from the scene to the acquired image, by avoiding at the same time the degradation due to an excessive aliasing distortion. All the scores are achieved by running the standard lossless compression scheme CCSDS 1.2.3.0-B-1 for multispectral/hyperspectral data, as a function of the cut-off value and different noise acquisition levels. The final results, and related plots, show that this analysis can suggest a suitable choice for the cut-off value, to ensure both a sufficient quality and low bit rates for the transmitted data to the ground station.
Highlights
In this paper, the real-time lossless compressibility of the acquired raw data in a satellite scenario is investigated when changes occur in the modulation transfer function (MTF) of the sensor optical system
Lossless algorithm ensure the recovering of the original data without any distortion, even if compression ratios (CR) of more of 3 are hardly achievable, because of the degradation relating to several impairments, which are caused by the presence of both internal and external noise sources [3], as acquisition noise, striping, registration errors, cosmic ray interactions, and so on
The reported experiments on the simulated raw data have been divided in two groups: (a) lossless bit rate (BR) of the reduced Yellowstone scene 0, by varying the image arrangement (BSQ or BIP), the cut-off value Kmin of the Gaussian MTF filter, the downsampling factor M, and the standard deviation σn of the signal-independent noise contribution; (b) RMSE versus BR for the Gaussian MTF filtered images in comparison with the image filtered by the 23-taps polynomial kernel, by varying the image arrangement (BSQ or BIP), the cut-off value Kmin, the standard deviation σn, and the scene under investigation
Summary
The real-time lossless compressibility of the acquired raw data in a satellite scenario is investigated when changes occur in the MTF of the sensor optical system. To well cope with designers’ needs, an algorithm must be suboptimal with respect to the state-of-art methods, in particular by resorting to integer arithmetics and to a high redundancy, so that the effects of possible transmission errors can be minimized. Such a factor is on behalf of lossless methods, which are more insensitive to the channel noise, because there is no superimposition between transmission errors and those due to the algorithm itself
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
